28
ログイン
言語:

WEKO3
  • トップ
  • コミュニティ
  • ランキング
AND
To

Field does not validate



インデックスリンク

インデックスツリー

メールアドレスを入力してください。

WEKO

One fine body…

WEKO

One fine body…

Item

{"_buckets": {"deposit": "ea3eca98-475e-4e02-bb39-1eb3dee224d2"}, "_deposit": {"id": "8565", "owners": [], "pid": {"revision_id": 0, "type": "depid", "value": "8565"}, "status": "published"}, "_oai": {"id": "oai:nagoya.repo.nii.ac.jp:00008565"}, "item_10_alternative_title_19": {"attribute_name": "\u305d\u306e\u4ed6\u306e\u8a00\u8a9e\u306e\u30bf\u30a4\u30c8\u30eb", "attribute_value_mlt": [{"subitem_alternative_title": "A Branch Prediction Scheme that Reduces Destructive Aliasing Using Branch Direction Bias"}]}, "item_10_biblio_info_6": {"attribute_name": "\u66f8\u8a8c\u60c5\u5831", "attribute_value_mlt": [{"bibliographicIssueDates": {"bibliographicIssueDate": "1999-05", "bibliographicIssueDateType": "Issued"}, "bibliographicIssueNumber": "5", "bibliographicPageEnd": "2131", "bibliographicPageStart": "2119", "bibliographicVolumeNumber": "40", "bibliographic_titles": [{"bibliographic_title": "\u60c5\u5831\u51e6\u7406\u5b66\u4f1a\u8ad6\u6587\u8a8c"}]}]}, "item_10_description_4": {"attribute_name": "\u6284\u9332", "attribute_value_mlt": [{"subitem_description": "\u547d\u4ee4\u30ec\u30d9\u30eb\u4e26\u5217\u6027\u3092\u5229\u7528\u3059\u308b\u9ad8\u6027\u80fd\u30d7\u30ed\u30bb\u30c3\u30b5\u306b\u304a\u3044\u3066\u306f, \u5206\u5c90\u4e88\u6e2c\u6a5f\u69cb\u306e\u4e88\u6e2c\u7cbe\u5ea6\u306f\u6027\u80fd\u306b\u5927\u304d\u306a\u5f71\u97ff\u3092\u4e0e\u3048\u308b. \u5206\u5c90\u4e88\u7cbe\u5ea6\u3092\u5236\u9650\u3059\u308b\u4e3b\u305f\u308b\u8981\u56e0\u306f, \u7af6\u5408\u3067\u3042\u308b\u3053\u3068\u304c\u77e5\u3089\u308c\u3066\u3044\u308b. \u3053\u308c\u307e\u3067\u7af6\u5408\u3078\u306e\u5bfe\u7b56\u3068\u3057\u3066, \u305d\u306e\u767a\u751f\u305d\u306e\u3082\u306e\u3092\u6e1b\u5c11\u3055\u305b\u308b\u8a66\u307f\u304c\u306a\u3055\u308c\u3066\u304d\u305f. \u3057\u304b\u3057\u3053\u306e\u65b9\u7b56\u3060\u3051\u3067\u306f\u4e88\u6e2c\u7cbe\u5ea6\u3092\u6539\u5584\u3059\u308b\u306b\u306f\u9650\u754c\u304c\u3042\u308b. \u3053\u308c\u306b\u5bfe\u3057\u3066\u6211\u3005\u306f, \u7af6\u5408\u306e\u4e2d\u3067\u3082\u4e88\u6e2c\u306b\u60aa\u5f71\u97ff\u3092\u4e0e\u3048\u308b\u7834\u58ca\u7684\u306a\u7af6\u5408\u306e\u524a\u6e1b\u306b\u6ce8\u76ee\u3057, \u7af6\u5408\u304c\u767a\u751f\u3057\u3066\u3082\u4e88\u6e2c\u7cbe\u5ea6\u3092\u5927\u304d\u304f\u4f4e\u4e0b\u3055\u305b\u306a\u3044\u30a2\u30d7\u30ed\u30fc\u30c1\u3092\u3068\u308b. \u3053\u306e\u30a2\u30d7\u30ed\u30fc\u30c1\u306b\u57fa\u3065\u304dsgshare\u3068\u547c\u3076\u65b0\u3057\u3044\u4e88\u6e2c\u6a5f\u69cb\u3092\u63d0\u6848\u3059\u308b. sgshare\u3067\u306f, \u30d1\u30bf\u30fc\u30f3\u5c65\u6b74\u30c6\u30fc\u30d6\u30eb\u30922\u3064\u306b\u5206\u96e2\u3057, \u5206\u5c90\u65b9\u5411\u306e\u504f\u308a\u304c\u540c\u4e00\u306e\u5206\u5c90\u3092\u4e00\u65b9\u306e\u30c6\u30fc\u30d6\u30eb\u306b\u306e\u307f\u30de\u30c3\u30d4\u30f3\u30b0\u3059\u308b\u3053\u3068\u306b\u3088\u308a, \u7834\u58ca\u7684\u306a\u7af6\u5408\u3092\u7121\u5bb3\u306a\u7af6\u5408\u306b\u5909\u63db\u3059\u308b. IBS-Ultrix\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3092\u7528\u3044\u305f\u30b7\u30df\u30e5\u30ec\u30fc\u30b7\u30e7\u30f3\u306b\u3088\u308a\u8a55\u4fa1\u3092\u3057\u305f\u7d50\u679c, sgshare\u306f\u5f93\u6765\u306egshare\u3068\u6bd4\u8f03\u3057\u3066, \u7d048K\u30d0\u30a4\u30c8\u306e\u30cf\u30fc\u30c9\u30a6\u30a7\u30a2\u30fb\u30b3\u30b9\u30c8\u306b\u304a\u3044\u3066, 0.05\uff5e0.89%\u4e88\u6e2c\u7cbe\u5ea6\u304c\u5411\u4e0a\u3059\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3057\u305f. \u307e\u305f, \u3053\u306e\u4e88\u6e2c\u7cbe\u5ea6\u306e\u6539\u5584\u306b\u3088\u308a, \u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u6027\u80fd\u306f\u4eca\u65e5\u306e\u30b9\u30fc\u30d1\u30fc\u30b9\u30ab\u30e9\u30fb\u30de\u30b7\u30f3\u306b\u304a\u3044\u3066\u5e73\u57477.19%, \u3088\u308a\u5e83\u3044\u547d\u4ee4\u767a\u884c\u5e45\u3068\u3088\u308a\u6df1\u3044\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3092\u6301\u3064\u5c06\u6765\u306e\u30b9\u30fc\u30d1\u30fc\u30b9\u30ab\u30e9\u30fb\u30de\u30b7\u30f3\u306b\u304a\u3044\u3066\u5e73\u574714.6%\u901f\u5ea6\u304c\u5411\u4e0a\u3059\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3057\u305f. Branch prediction accuracy makes a tremendous impact on modern high-performance microprocessors that exploit instruction-level parallelism. Aliasing is a known major factor that limits prediction accuracy. A several studies have attempted to reduce the total rate of aliasing. Yet, only this approach is not enough for further improvement of prediction accuracy. In this paper, we focus on reducing destructive aliasing that has a negative impact on prediction accuracy. We propose a new branch prediction scheme, called sgshare, which modestly affects prediction accuracy even when aliasing occurs. Sgshare divides the pattern history table into two parts, and maps branches with the same direction bias to only a single part of the table. This mechanism converts destructive aliasing into harmless aliasing. Our evaluation results show that sgshare achieves 0.05-0.89% better prediction accuracy than the conventional gshare with approximately 8 K-byte hardware cost. This improvement of the prediction accuracy contributes to the performance of a current superscalar processor by 7.19%, and that of a future superscalar processor with a wider instruction issue and a deep pipeline by 14.6%.", "subitem_description_type": "Abstract"}]}, "item_10_identifier_60": {"attribute_name": "URI", "attribute_value_mlt": [{"subitem_identifier_type": "HDL", "subitem_identifier_uri": "http://hdl.handle.net/2237/10314"}]}, "item_10_publisher_32": {"attribute_name": "\u51fa\u7248\u8005", "attribute_value_mlt": [{"subitem_publisher": "\u60c5\u5831\u51e6\u7406\u5b66\u4f1a"}]}, "item_10_select_15": {"attribute_name": "\u8457\u8005\u7248\u30d5\u30e9\u30b0", "attribute_value_mlt": [{"subitem_select_item": "publisher"}]}, "item_10_source_id_7": {"attribute_name": "ISSN", "attribute_value_mlt": [{"subitem_source_identifier": "03875806", "subitem_source_identifier_type": "ISSN"}]}, "item_10_text_14": {"attribute_name": "\u30d5\u30a9\u30fc\u30de\u30c3\u30c8", "attribute_value_mlt": [{"subitem_text_value": "application/pdf"}]}, "item_creator": {"attribute_name": "\u8457\u8005", "attribute_type": "creator", "attribute_value_mlt": [{"creatorNames": [{"creatorName": "\u91ce\u53e3, \u826f\u592a"}], "nameIdentifiers": [{"nameIdentifier": "24101", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "NOGUCHI, RYOTA"}], "nameIdentifiers": [{"nameIdentifier": "24102", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "\u68ee, \u6566\u53f8"}], "nameIdentifiers": [{"nameIdentifier": "24103", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "MORI, ATSUSHI"}], "nameIdentifiers": [{"nameIdentifier": "24104", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "\u5c0f\u6797, \u826f\u592a\u90ce"}], "nameIdentifiers": [{"nameIdentifier": "24105", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "KOBAYASHI, RYOTARO"}], "nameIdentifiers": [{"nameIdentifier": "24106", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "\u5b89\u85e4, \u79c0\u6a39"}], "nameIdentifiers": [{"nameIdentifier": "24107", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "ANDO, HIDEKI"}], "nameIdentifiers": [{"nameIdentifier": "24108", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "\u5cf6\u7530, \u4fca\u592b"}], "nameIdentifiers": [{"nameIdentifier": "24109", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "SHIMADA, TOSHIO"}], "nameIdentifiers": [{"nameIdentifier": "24110", "nameIdentifierScheme": "WEKO"}]}]}, "item_files": {"attribute_name": "\u30d5\u30a1\u30a4\u30eb\u60c5\u5831", "attribute_type": "file", "attribute_value_mlt": [{"accessrole": "open_date", "date": [{"dateType": "Available", "dateValue": "2018-02-19"}], "displaytype": "detail", "download_preview_message": "", "file_order": 0, "filename": "40-5-2119-2131.pdf", "filesize": [{"value": "1.2 MB"}], "format": "application/pdf", "future_date_message": "", "is_thumbnail": false, "licensetype": "license_free", "mimetype": "application/pdf", "size": 1200000.0, "url": {"label": "40-5-2119-2131.pdf", "url": "https://nagoya.repo.nii.ac.jp/record/8565/files/40-5-2119-2131.pdf"}, "version_id": "095e0f41-20c8-49f9-b6aa-a6711d06a4c1"}]}, "item_language": {"attribute_name": "\u8a00\u8a9e", "attribute_value_mlt": [{"subitem_language": "jpn"}]}, "item_resource_type": {"attribute_name": "\u8cc7\u6e90\u30bf\u30a4\u30d7", "attribute_value_mlt": [{"resourcetype": "journal article", "resourceuri": "http://purl.org/coar/resource_type/c_6501"}]}, "item_title": "\u5206\u5c90\u65b9\u5411\u306e\u504f\u308a\u3092\u5229\u7528\u3057\u7834\u58ca\u7684\u7af6\u5408\u3092\u4f4e\u6e1b\u3059\u308b\u5206\u5c90\u4e88\u6e2c\u65b9\u5f0f", "item_titles": {"attribute_name": "\u30bf\u30a4\u30c8\u30eb", "attribute_value_mlt": [{"subitem_title": "\u5206\u5c90\u65b9\u5411\u306e\u504f\u308a\u3092\u5229\u7528\u3057\u7834\u58ca\u7684\u7af6\u5408\u3092\u4f4e\u6e1b\u3059\u308b\u5206\u5c90\u4e88\u6e2c\u65b9\u5f0f"}]}, "item_type_id": "10", "owner": "1", "path": ["320/321/322"], "permalink_uri": "http://hdl.handle.net/2237/10314", "pubdate": {"attribute_name": "\u516c\u958b\u65e5", "attribute_name_i18n": "\u516c\u958b\u65e5", "attribute_value": "2008-08-04"}, "publish_date": "2008-08-04", "publish_status": "0", "recid": "8565", "relation": {}, "relation_version_is_last": true, "title": ["\u5206\u5c90\u65b9\u5411\u306e\u504f\u308a\u3092\u5229\u7528\u3057\u7834\u58ca\u7684\u7af6\u5408\u3092\u4f4e\u6e1b\u3059\u308b\u5206\u5c90\u4e88\u6e2c\u65b9\u5f0f"], "weko_shared_id": 3}
  1. B200 工学部／工学研究科
  2. B200a 雑誌掲載論文
  3. 学術雑誌

分岐方向の偏りを利用し破壊的競合を低減する分岐予測方式

http://hdl.handle.net/2237/10314
716dd68f-3781-40a6-8992-25ffd1d70d82
Name / File License Actions
40-5-2119-2131.pdf 40-5-2119-2131.pdf (1.2 MB)
item type 学術雑誌論文 / Journal Article(1)
公開日 2008-08-04
タイトル
タイトル 分岐方向の偏りを利用し破壊的競合を低減する分岐予測方式
その他のタイトル
その他のタイトル A Branch Prediction Scheme that Reduces Destructive Aliasing Using Branch Direction Bias
著者 野口, 良太

× 野口, 良太

WEKO 24101

野口, 良太
Search repository
NOGUCHI, RYOTA

× NOGUCHI, RYOTA

WEKO 24102

NOGUCHI, RYOTA
Search repository
森, 敦司

× 森, 敦司

WEKO 24103

森, 敦司
Search repository
MORI, ATSUSHI

× MORI, ATSUSHI

WEKO 24104

MORI, ATSUSHI
Search repository
小林, 良太郎

× 小林, 良太郎

WEKO 24105

小林, 良太郎
Search repository
KOBAYASHI, RYOTARO

× KOBAYASHI, RYOTARO

WEKO 24106

KOBAYASHI, RYOTARO
Search repository
安藤, 秀樹

× 安藤, 秀樹

WEKO 24107

安藤, 秀樹
Search repository
ANDO, HIDEKI

× ANDO, HIDEKI

WEKO 24108

ANDO, HIDEKI
Search repository
島田, 俊夫

× 島田, 俊夫

WEKO 24109

島田, 俊夫
Search repository
SHIMADA, TOSHIO

× SHIMADA, TOSHIO

WEKO 24110

SHIMADA, TOSHIO
Search repository
抄録
内容記述タイプ Abstract
内容記述 命令レベル並列性を利用する高性能プロセッサにおいては, 分岐予測機構の予測精度は性能に大きな影響を与える. 分岐予精度を制限する主たる要因は, 競合であることが知られている. これまで競合への対策として, その発生そのものを減少させる試みがなされてきた. しかしこの方策だけでは予測精度を改善するには限界がある. これに対して我々は, 競合の中でも予測に悪影響を与える破壊的な競合の削減に注目し, 競合が発生しても予測精度を大きく低下させないアプローチをとる. このアプローチに基づきsgshareと呼ぶ新しい予測機構を提案する. sgshareでは, パターン履歴テーブルを2つに分離し, 分岐方向の偏りが同一の分岐を一方のテーブルにのみマッピングすることにより, 破壊的な競合を無害な競合に変換する. IBS-Ultrixベンチマークを用いたシミュレーションにより評価をした結果, sgshareは従来のgshareと比較して, 約8Kバイトのハードウェア・コストにおいて, 0.05～0.89%予測精度が向上することを確認した. また, この予測精度の改善により, プロセッサの性能は今日のスーパースカラ・マシンにおいて平均7.19%, より広い命令発行幅とより深いパイプラインを持つ将来のスーパースカラ・マシンにおいて平均14.6%速度が向上することを確認した. Branch prediction accuracy makes a tremendous impact on modern high-performance microprocessors that exploit instruction-level parallelism. Aliasing is a known major factor that limits prediction accuracy. A several studies have attempted to reduce the total rate of aliasing. Yet, only this approach is not enough for further improvement of prediction accuracy. In this paper, we focus on reducing destructive aliasing that has a negative impact on prediction accuracy. We propose a new branch prediction scheme, called sgshare, which modestly affects prediction accuracy even when aliasing occurs. Sgshare divides the pattern history table into two parts, and maps branches with the same direction bias to only a single part of the table. This mechanism converts destructive aliasing into harmless aliasing. Our evaluation results show that sgshare achieves 0.05-0.89% better prediction accuracy than the conventional gshare with approximately 8 K-byte hardware cost. This improvement of the prediction accuracy contributes to the performance of a current superscalar processor by 7.19%, and that of a future superscalar processor with a wider instruction issue and a deep pipeline by 14.6%.
出版者
出版者 情報処理学会
言語
言語 jpn
資源タイプ
資源 http://purl.org/coar/resource_type/c_6501
タイプ journal article
ISSN
収録物識別子タイプ ISSN
収録物識別子 03875806
書誌情報 情報処理学会論文誌

巻 40, 号 5, p. 2119-2131, 発行日 1999-05
フォーマット
application/pdf
著者版フラグ
値 publisher
URI
識別子 http://hdl.handle.net/2237/10314
識別子タイプ HDL
Back
0
views
See details
Views

Versions

Ver.1 2021-03-01 12:08:10.437213
Show All versions

Share

Mendeley CiteULike Twitter Facebook Print Addthis

Cite as

Export

OAI-PMH
  • OAI-PMH JPCOAR
  • OAI-PMH DublinCore
  • OAI-PMH DDI
Other Formats
  • JSON
  • BIBTEX

Confirm

Powered by CERN Data Centre & Invenio